def find_links(current_text, parent_text, dh_finder, *args):
"""
:param current_text: Text being parsed. Must have keys "name" and "text"
:param parent_text: Parent text text should be linked to. Must have keys "name" and "text"
:param dh_finder: function to find dh in text
:return: List of links
"""
matcher = Match(guess=True)
links = []
for chapter_num, chapter in enumerate(current_text['text']):
if chapter:
dh_list = [
dh_finder(seif[0], *args) for seif in chapter
if dh_finder(seif[0], *args)
]
matches = matcher.match_list(dh_list,
parent_text['text'][chapter_num],
parent_text['name'])
links.append(
build_links(parent_text['name'], current_text['name'],
chapter_num + 1, matches))
return [linker for sublist in links for linker in sublist]
def find_links(current_text, parent_text, dh_finder, *args):
"""
:param current_text: Text being parsed. Must have keys "name" and "text"
:param parent_text: Parent text text should be linked to. Must have keys "name" and "text"
:param dh_finder: function to find dh in text
:return: List of links
"""
matcher = Match(guess=True)
links = []
for chapter_num, chapter in enumerate(current_text['text']):
if chapter:
dh_list = [dh_finder(seif[0], *args) for seif in chapter if dh_finder(seif[0], *args)]
matches = matcher.match_list(dh_list, parent_text['text'][chapter_num], parent_text['name'])
links.append(build_links(parent_text['name'], current_text['name'], chapter_num+1, matches))
return [linker for sublist in links for linker in sublist]
def create_links(sanhedrin_ja, yad_ramah_ja):
list_of_links = []
amud_number = 1
match_object = Match(in_order=True,
min_ratio=80,
guess=False,
range=False,
can_expand=True)
for amud_of_sanhedrin, amud_yad_ramah in zip(sanhedrin_ja, yad_ramah_ja):
ref = 'Sanhedrin {}'.format(AddressTalmud.toStr('en', amud_number))
the_first_few_words = take_the_first_few_words_of_each_paragraph(
amud_yad_ramah)
matches_dict = match_object.match_list(the_first_few_words,
amud_of_sanhedrin, ref)
for key in matches_dict:
for match in matches_dict[key]:
if match != 0:
# print'Amud: {} comment: {} corresponds to {}'.format(AddressTalmud.toStr('en', amud_number), key, match)
print create_link_text(amud_number, match, key)
list_of_links.append(
create_link_text(amud_number, match, key))
amud_number += 1
return list_of_links